> -----Original Message----- > From: Jan Beulich [mailto:jbeul...@suse.com] > Sent: 26 September 2018 12:57 > To: Paul Durrant <paul.durr...@citrix.com> > Cc: Julien Grall <julien.gr...@arm.com>; Andrew Cooper > <andrew.coop...@citrix.com>; Roger Pau Monne <roger....@citrix.com>; > Stefano Stabellini <sstabell...@kernel.org>; xen-devel <xen- > de...@lists.xenproject.org> > Subject: RE: IOREQ server on Arm > > >>> On 26.09.18 at 13:02, <paul.durr...@citrix.com> wrote: > > --- a/xen/common/memory.c > > +++ b/xen/common/memory.c > > @@ -1105,8 +1105,11 @@ static int acquire_resource( > > > > for ( i = 0; !rc && i < xmar.nr_frames; i++ ) > > { > > - rc = set_foreign_p2m_entry(currd, gfn_list[i], > > - _mfn(mfn_list[i])); > > + rc = (xmar.flags & XENMEM_rsrc_acq_caller_owned) ? > > + guest_physmap_add_entry(currd, gfn_list[i], > > + _mfn(mfn_list[i]), 0, > p2m_ram_rw) : > > + set_foreign_p2m_entry(currd, gfn_list[i], > > + _mfn(mfn_list[i])); > > /* rc should be -EIO for any iteration other than the first > */ > > if ( rc && i ) > > rc = -EIO; > > > > But the guest_physmap_add_entry() is problematic as it will IOMMU map > pages > > as well, which is probably not wanted. > > Yeah, I'd prefer if we avoided establishing IOMMU mappings here. > How about transforming set_foreign_p2m_entry() into > set_special_p2m_entry(), with a type passed in? >
That sounds like it might work. Julien, do you want page types to distinguish caller-owned resources from normal RAM are you ok with p2m_ram_rw even though it could be subject of another domain's foreign map? Paul > Jan > _______________________________________________ Xen-devel mailing list Xen-devel@lists.xenproject.org https://lists.xenproject.org/mailman/listinfo/xen-devel
